- BugFix: allow tested_reference_str to be None - #10834
Conversation
47328b5 to
f0776ff
Compare
There was a problem hiding this comment.
We would like to avoid that new argument. We've been discussing and we might prefer if self.requires(None) and self.build_requires(None) don't do anything, that is actually more accurate to the current behavior with 1.X, because the whole graph is not computed either, just the test_package and its requires. Probably we need to initialize the self.tested_reference_str to None always. WDYT?
I am personally fine with that. it will allow to compute requirements of test_package in our CI, and we already have requirements of the package itself computed. (well, it might not work if test_package changes options of requirements, but it doesn't matter for CCI in this case, as it will fail for a different reason - missing package, cause package built is not one being tested). |
f0776ff to
9cd2470
Compare
|
@lasote I've changed the implementation according to your suggestion. |
| .with_test_type("explicit")\ | ||
| .with_requirement("placeholder")\ | ||
| .with_build_requirement("placeholder") | ||
| conanfile = str(conanfile).replace('"placeholder"', 'self.tested_reference_str') |
There was a problem hiding this comment.
Not worth the trick. Just write a explicit conanfile it will be just a couple of lines longer, but way more explicit and understandable.
Signed-off-by: SSE4 <tomskside@gmail.com>
9cd2470 to
dbb7fa8
Compare
closes: #10714
Changelog: Feature: Allow
tested_reference_strto beNone.Docs: omit
developbranch, documenting this one.Note: By default this PR will skip the slower tests and will use a limited set of python versions. Check here how to increase the testing level by writing some tags in the current PR body text.